Reprint of paper, presented at the International Seminar on Steel

Structures in Fire, Tongji University, Shanghai, P.R. China,

November 1-3, 2001.
See note on page 7 for adjustment to final version of EC3-1.2.

THE NEW EUROCODE ON FIRE DESIGN OF STEEL STRUCTURES

Leen Twilt
Convenor of the Project Team on the conversion of ENV1993-1.2 to prEN1993-1.2
TNO, The Netherlands

ABSTRACT
In 1995 the part of Eurocode 3: Design of Steel Structures dealing with the fire design was published as
a voluntary European standard: ENV1993-1.2. By the middle of 1999, work has started to convert this
ENV into a harmonized technical specification (EN). Basis of the work is formed by the comments
received on the ENV version from the various EU member states. A substantial part of the comments
refers to technical issues, other comments are of an editorial nature. This paper concentrates on the
technical comments. As a result of these technical comments, various modifications of ENV1993-1.2
which are proposed. Important topics in this respect are:
deformation criteria under fire conditions
material properties at elevated temperature
fire design of connections
classifcation of cross sections at elevated temperatures
buckling curves and M-N interaction at elevated temperatures
effect of non uniform temperature distribution on the mechanical resistance of beams
calculation rules for thermal response of bare steel beams
The modifications regarding each of the above topics will globally be reviewed. For details, refer to the
final draft of prEN1993-1.2 which is available by now and to background documents. References to these
background documents are included in the paper.

KEYWORDS
Eurocode, steel structures, fire, thermal response, mechanical response, deformation criteria, shadow
effect, buckling curves, M-N interaction, non uniform temperature distribution

1.

INTRODUCTION

Part 1.2 of Eurcode 3 deals with the fire design of steel structures and was issued in September 1995 [1].
This part was, as all the other parts of the Eurocodes, published as an ENV, i.e. application within the
European Union is on a voluntary basis only. The various EU Member States can, by means of so-called
National Application Documents (NADs), specify modifications, which then hold within the legislative
system of that Member State. Many EU Member States have used this opportunity. Hence, the aim of the
Eurocodes, i.e. harmonizing the rules for structural (fire) design is far from being achieved by their
introduction.
By the middle 1999, the activities have started to convert the present Eurocodes from ENVs into ENs.
This work is under responsibility of CEN TC250 and is more in particular guided by various Sub
Committees, which have delegated the actual drafting activities to Project Teams. For the design of steel
structures Sub Committee 3 (SC3) is in charge; preparation of the conversion text of the fire part
(prEN1993-1.2) is undertaken by Project Team 2 (SC3/PT2)1.
In the period between mid 1995 and mid 1999, practical experience has been obtained with the
application of EC3-1.2. On the basis of this experience, 207 comments have been received from 19
different Member Sates and international branch organizations. About half of the comments refer to
technical aspects of the ENV version of EC3-1.2; the other half to editorial matters. For the distribution of
the technical comments over the main issues of the EC3-1.2, refer to Fig. 1.
120

100

80

number

Overall comments
Scope & definitions
Basic principle & rules
Material propeties
Structural fire design
Annexes

60

40

20

0
1

main issue

Figure 1: Comments per main issue

In this figure a separate category has been included for comments which are not related to one single
aspect, the so-called overall comments. One may conclude from Fig. 1, that by far most comments refer
to the rules, specifically given for the Structural Fire Design (chapter 4 of ENV1993-1.2). Within this
category, most attention (i.e. 85 %) is on the so-called simple calculation rules (section 4.2 of
ENV1993-1.2). Important topics in this category are:
connections;
classification of cross sections;
1

Membership of SC3/PT2: N. Andersen (DK), M. Fontana (CH), J-M. Franssen (B), D. Moore (UK), G. Sedlacek/
C. Heinemeyer secretariat - (D), L. Twilt convenor (NL).

buckling curves and M-N interaction;

effect of non uniform temperature distribution on the mechanical resistance of beams (1-factor);
calculation rule for the thermal response of bare steel beams.

Topics of a more general nature, which have been strongly commented upon, include:
deformation criteria;
material properties.
In this paper, the main modifications of the ENV version of EC3-1. will be discussed in a global manner.
For details, refer to the draft of the EN version [2], which will become available by the end of 2001 and to
the accompanying background documents. In the text of the paper reference will be made to these
background documents, where appropriate.
The discussion on the above topics is embedded in the structure chosen for EC3-1.2:
General, dealing with formal aspects such as scope, definitions etc.
Basic principles and rules, dealing with more general aspects such as deformation criteria etc;
Material properties, both thermal and mechanical;
Simple calculation models, as main stream for the structural fire design, dealing with component
design (beams, columns etc.)
The above subdivision corresponds - by and large - with the various sections of EC3-1.2. With a view to
provide some overall insight in the set up of EC3-1.2, the intent of each of the sections will be reviewed
in general terms, before commenting on the anticipated modifications resulting from the conversion
process

2.

GENERAL

Section 1 of EC3-1.2 (General) contains:

scope
normative references
basic assumption
definitions
symbols
Most of the clauses given are so-called model clauses, i.e. they are identical in the various fire parts of
Eurocode (concrete, steel, timber etc.). The Horizontal Group Fire (HGF)2 is responsible for harmonizing
the model clauses. As far as EC3-1.2 is concerned, the main deviation from the model clauses is due to
the fact that for the fire design of steel structures the separating function is not taken into account.
Compared to the ENV version of EC3-1.2 no significant modifications are forseen.

Membership HGF: J. Kruppa convenor - (F), together with the convenors of the various Project Teams, responsible for
the conversion of the Eurcode parts Fire

3.

BASIC PRINCIPLES AND RULES

Section 2 of EC3-1.2 (Basic principles and rules) contains:

performance requirements;
actions;
design values of material properties;
assessment methods.
Again, most of the clauses in this chapter are model clauses, i.e. clauses which are identical for all the
fire parts of the Eurocode (see paragraph 2 above).
The main modification in the EN compared to the ENV version concerns the deformation criteria. In the
ENV version, a check on deformations was required for steel members with fire insulation and/or for
members, the deformation of which might endanger the functioning of adjacent separating elements (e.g.
partitions), if the deformation of the load bearing structure requires consideration. When using in
simple calculation models - a reduced strength reduction factor (kx,, based on a strain of 0.5%), it is not
necessary to check the deformation criteria explicitly.
From the comments received, it became clear that there was major concern with this approach. Important
arguments against were (1) the open end formulation (2) lack of functional background and (3)
exceptional position for steel, since for other structural materials (concrete, timber etc.) such rules do not
exist. Therefore, SC3-PT2 paid due attention to the deformation criteria. As a first step, a more
functional definition for kx, was derived, based on a maximum strain 1% [3]. In Fig. 2, the strength
reduction factor ky,, to be applied if deformation criteria need not to be satisfied as well as the new kx,
factor are presented, together with the kx, as specified in ENV1993-1.2.
1.2
1

k-value

0.8
k(y)
k(x)-new
k(x)-old

0.6
0.4
0.2
0
0 100 200 300 400 500 600 700 800 900 100 110
0
0

steel temperature [C]

Figure 2: Strength reduction factors as function of the steel temperature

It can be shown that the effect of changing from ky, (no deformation criteria to be satisfied) to kx,-new
(deformation criteria to be satisfied) in terms of fire resistance time is only marginal [4]. For this reason
and also considering the other arguments mentioned above it has been decided to refrain from a check
on deformation if the separating elements and/or the protection system (if any) are designed according to
the CEN classification system & rules.
The other modifications mainly concern stream lining of the text and do not actually have a big
technical impact. This means, for example, that the EN version of EC3-1.2 still offers three levels for
assessing the structural behaviour of fire exposed steelwork:
member analysis;
analysis of the part of the structure;
global structural analysis.
However, in view of the fact that member analysis is most often used in practice, emphasis is on this
option.

4.

MATERIAL PROPERTIES

In section 1 of EC3-1.2 both the mechanical and thermal properties of steel to be used in the structural
fire safety design are specified. As far as the properties of the fire protection materials are involved,
reference is made to the relevant CEN standard [5] (as in the ENV version of EC3-1.2).
New - compared to the ENV version of EC3-1.2 - is that the mechanical properties are now not only
given for carbon steel, but for stainless steel as well. The latter information is presented in a separate
Annex and is based on an extensive European research project, which was recently completed [6].
The mechanical properties of carbon steel are the same as specified earlier in the ENV version of EC31.2. In here, the factors for strength reduction and Youngs modulus reduction are presented by both
tabulated data (derived from tests) and equations. The latter are approximations of the tabulated data and
consequently there is no exact match. This is illustrated in fig. 3, where the error, when using the equation
is plotted against the steel temperature. Especially for steel temperatures beyond 700 C the error is rather
large. Since unambiguous formulations are not allowed in code text, it has been decided to delete the
equations. The strength reduction factor for satisfying deformation criteria is still there, however as
explained earlier on an informative basis only.

15.0
10.0

error [%}

5.0
0.0
-5.0 0

200

400

600

800

1000

1200

error

-10.0
-15.0
-20.0
-25.0
steel temperature [C]

Figure 3: Error when for the stress reduction factor the equation according to [1] is
used in stead of the tabulated data
Also the thermal properties of carbon steel are unchanged compared the ENV version of EC3-1.2.
However, the time independent design values for the thermal elongation, the specific heat and the thermal
conductivity have been deleted, since the widespread use of computers makes this simplification
unnecessary.

5.

SIMPLE CALCULATION MODELS

5.1

Introduction

The simple calculation models are dealt with in section 4.2 of EC3-1.2 (Structural fire design), and form
by far the most extensive part of this code. Prerequisite for the simple calculation models is standard fire
exposure and member analysis. The following topics are covered:
general guidance, e.g. with respect to the fire design of connections;
classification of cross-sections
mechanical resistance of
tension members;
compression members (cross section classes 1, 2, 3);
beams, including the effects of non-uniform temperature distribution (cross section classes 1,
2, 3);
members under combined bending and axial compression (cross section classes 1, 2, 3);
members with cross section class 4;
steel temperature development for:
unprotected internal steelwork;
internal steelwork insulated by fire protection material;
internal steelwork that is protected by heat screens;
external steelwork.

Topics which are subject to substantial modifications compared to the ENV version of EC3-1.2 are
printed in bold in the above list. They will briefly be discussed hereafter.
5.2

Fire design of connections

The ENV version of EC3-1.2 specifies that the resistance of connections need not to be checked, provided
the thermal resistance of the fire protection of the connection is not less than the minimum value of the
thermal resistance of the fire protection of any of the steel members joint to that connection. The thermal
resistances are defined by the ratio between the appropriate values of the insulation thickness (df) and the
thermal conductivity (f).
In the EN version this rule is - for reasons of consistency - extended with an additional condition:
utilization should be less than the maximum value of any of the connected members. As a simplification,
the comparison of the level of utilization within a connection may be performed as for room temperature.
As an alternative for the fire design based on the above simple rules, the fire resistance of connections
may be determined using a method specified in Annex D (informative only) of the EN version. This
calculation method holds both for bolted (shear/tension) and welded connections (butt & fillet welds) and
is based on an assessment of the mechanical resistance of the connection and its temperature response,
when exposed to fire. The method is based on [7].
5.3

Classification of cross sections

In the ENV version of EC3-1.2 cross sections of beams are classified by means of the parameter:
= [(235/fy)(kE, / ky, )]0.5
with:
fy
ky, , kE,

is
is

yield stress at room temperature;

strength reduction and Youngs modulus reduction factor respectively.

The parameter was simply obtained by substituting elevated temperature values for yield strength and
Youngs modulus in the equations used for room temperature classification of the cross section. The
variation of (kE, / ky, )0.5 as function of temperature shows an extremely irregular behaviour. See Fig. 4.
As a result, the ENV classification rule gives strange result: starting with class 1, a cross section can
move into class 2 or 3 and back. Since there is no evidence on the accuracy of the ENV rule and also the application of such a rule hampers the practical design procedure significantly, SC3/PT2 proposes to
retain classification as in cold design, i.e. independent of the temperature.3

Note: Further to discussions within SC3, borderline values for b/t are based on = 0,85(235/fy)0,5, thus creating a safety
margin for classifying cross sections of beams at elevated temperature. See EC3-1.2: 4.2.2 (1).

sqrt(kE/ky)
1.2

sqrt(KE/Ky)

1
0.8
0.6
0.4
0.2
0
0

200

400

600

800

1000

1200

temperature [C]

Figure 4 : Variation of the factor (kE, / ky, )0.5 as function of the steel temperature
5.4

Mechanical resistance of compression members

The present EC3-1.2 rules for the mechanical resistance of compression members (buckling curves at
elevated temperatures), are based on buckling curve c at room temperature (irrespective of the type of
cross section) in combination with a reduced yield stress, by applying the strength reduction factor ky, .
As input parameter, a non dimensional, temperature dependent slenderness ratio ( ) is used, following
from:

= [k y , / k E , ]0,5
with:

ky, , kE,

is
is

non dimensional slenderness ratio at room temperature

strength reduction and Youngs modulus reduction factor respectively

In the mean time extensive research, sponsored by the European Commission for Coal and Steel (ECSC)
has been carried out in this field [8], [9]. As a result, new buckling curves at elevated temperatures could
be established. In this approach, the dimensionless slenderness ratio, as specified in the ENV version of
EC3-1.2, is maintained, i.e. temperature dependent. An important advantage of the new rule is that it is
consistent with the room temperature design. Also, somewhat less conservative results are obtained, when
compared to the ENV version of EC3.1. Refer to Fig. 5 for a comparison between the old and the
new buckling curves. For background information, see [8].

1.0
0.9

Nu(T) / Npl(T)

0.8
0.7
0.6

New S500
New S355
New S235
ENV 1993

0.5
0.4
0.3
0.2
0.1
0.0

0.0
0.5
1.0
2.0
2.5
3.0
Figure 5: Comparison
between the
present 1.5
buckling curves
at elevated
temperature
and
Lamda
rel(Tcrit)
the new proposal

5.5

Mechanical resistance of beams with a non uniform temperature across the cross section of the
beam

When using simple calculation models for the determination of the mechanical resistance of beams, the
ENV version of EC3-1.2 offers the possibility to assume a uniform temperature distribution across the
cross section of the beam. Especially for three side exposed beams (e.g. beam with a concrete floor
above) the temperature distribution will in reality - be far from uniform. This is corrected for by
introducing an adaptation factor 1 smaller than unity, by which the resistance has to be divided4. The
value of 1 depends on the way of exposure. In ENV1993-1.2 the following is specified:
for a beam exposed on all four sides:
1 = 1.0
for a beam exposed on three sides, with a slab (concrete or composite) above:
1 = 0.7
A wide variety of comments have been received with regard to the 1-value. Suggested values vary from
0.7 (as in ENV1993-1.2) to 1.0 (completely ignoring the effect of the non-uniform temperature
distribution).
The Project Team has studied this issue in detail, basing itself on both experimental and theoretical work,
carried out in the past [10], [11], [12]. From this study, it became clear the experimental evidence
supporting a 1-value of 0.7 only exists for bare steel beams. See Fig. 6.

To account for a possible non-uniform temperature distribution along a beam, an adaptation factor 2 is introduced in EC31.2. Since 2 will not be modified as result of the conversion process, it will not be discussed here.

Test and computed beam performance

Reference steel temperature (C)

900
Test results
K = 1.00

800

K = 0.85
K = 0.70

700

600

500
0.0

0.1

0.2

0.3
0.4
Utilization

0.5

0.6

0.7

Figure 6: Test results and various options for the 1 factor for unprotected beams

For fire insulated steel beams, no such evidence is available. However, the results of theoretical studies
suggest for these cases a 1-value of 0.85 for practical load levels. See Fig. 7. For this reasons SC3-PT2
proposes to change the 1-values in the EN version of EC3-1.2 for 3 side exposed beams accordingly. For
4 side exposed beams the situation remains the same, i.e. 1 = 1.0. See also [13].

Figure 7: The 1-factor as function of the utilization factor for steel members protected by
contour insulation (calculation by means of an advanced computer model) [4]

5.6

Combined bending and axial compression

The rules on combined bending and axial compression (M-N interaction) included in the ENV version of
EC3-1.2 are based on research carried out in mid nineties [14]. There is no reason to doubt its validity
[14]. The rules are formulated in conjunction with the rules for room temperature conditions specified
part 1.1 of EC3 [15]. The complication now is that the latter rules will be modified, but that by the
middle of 2001 it was not yet known which of the available options would be chosen. In view of the
tough time schedule, and also for budgetary reasons, the Project Team has decided to stick to the room
temperature rules according to ENV 1993-1.1. Since these rules will not be available anymore in the EN
version of part 1.1, they will be incorporated in part 1.2 of the EN version. Note that the this modification
has no effect what so ever in terms of calculation results.
5.7

Calculation rules for temperature development in unprotected internal steelwork

Calculation rules for the temperature development in bare (i.e. unprotected) steelwork specified in the
ENV version of EC3-1.2 are based on conventional values for the coefficients of both the radiative and
convective heat transfer. These values are chosen such that reasonable agreement with test results is
obtained, leading however to assumptions which are from a physical point of view not very
convincing. This particularly holds for the radiative heat transfer, where rather unrealistic values for the
emissivity coefficients for both the steel surface and the fire environment have to be assumed: a value for
the resultant emissivity as low as 0.5 is necessary in order to achieve a reasonable match with test results.
This problem has become even more explicit when introducing the so-called Plate Thermometer (in stead
of the common thermocouples) as measuring device for controlling the gas temperature during standard
fire resistance testing [16], [17].
With a view to arrive at more realistic and consistent calculation rules for the temperature development in
bare steel members and also to stay in line with future standard fire resistance testing practice, the Project
Team proposes to use more realistic values for the emissivity coefficients of steel (a = 0.7; being a low,
but realistic value) and the fire environment (fi = 1.0; as direct consequence of using the Plate
Thermometer for furnace control [16]) .
The uplifting effect in terms of calculated temperatures of these modifications is by and large compensated by taking into consideration the so-called shadow effect, which is not explicitly taken into
account in the present calculation rules. Assuming fully embedded members (as in the case of simple
calculation models), the shadow effect is caused by local shielding of the radiation, due the shape of the
steel profile. It plays a role for profiles with a concave shape, such as I-sections; for profiles with a
convex shape, such as tubes, it does not exist (no local shielding).
The increase of temperature a,t in an unprotected steel member during a time interval t may then be
determined from:
/V
a,t= k shadow Am h& net,d t
ca a

with:
k shadow
Am / V

is
is

correction factor for the shadow effect

the section factor for unprotected steel members;

is
is
is
is

Am
V
ca
h& net,d

the surface area of the member per unit length;

the volume of the member per unit length;
the specific heat of steel [J/kgK];
the design value of the net heat flux per unit area, with a = 0.7 and fi = 1.0
[W/m2];
the time interval [seconds];
the unit mass of steel, from section 3 [kg/m3].

is
is

New in the expression compared to the ENV version of EC3-1.2 is the correction factor k shadow for
the shadow effect5. It can be shown that the shadow effect is reasonably well described by taking
kshadow = 0.9 [Am/V]box/[Am/V]
with
[Am/V]box

is

box value of the section factor

In Fig. 8, temperatures, calculated on the basis of the above new rule (including the shadow effect &
realistic values for the heat transfer coefficients) have been compared with values calculated on the basis
of the old rule (i.e. no shadow effect & heat transfer coefficients according to the ENV version of EC31.2). The temperatures have been calculated for various times of standard fire exposure and for all listed
European I sections (IPE, HEA, HEB etc.).

1.1

900
800

0.9

old, rule, t,i

700
600

t=10
t=20
t=30
t=60

500
400
300
200
100
0
0

100

200

300

400

500

600

700

800

900

new, rule,t,i [0C]

Figure 8: Steel temperatures in bare internal steelwork, calculated according to the present rule
of EC3-1.2, compared with calculation results according to the proposed new rule.

The correction factor makes no distinction between radiative and convective heat flux. It is clear that convective heat
transfer is less affected by a shadow effect than radiation is, This is ignored since convection plays only a minor role over
the heating period.

It follows from the definition of kshadow that for tube profiles, the shadow effect is not activated, since
[Am/V] = [Am/V]box
Note that for protected members, the shadow effect is less relevant since the effect of the fire insulation
overrules that of the heat transfer at the surface of the member. In EC3-1.2 the latter effect is for
protected members even completely ignored.

6.

CONCLUSIONS

Under responsibility of Sub Committee 3 of CEN TC250, Project Team 2 has worked on the conversion
of the Eurocode on the Fire Design of Steel Structures (ENV1993-1.2) from an ENV to an EN. In the
scope of this work more than 100 comments with regard to technical aspects of ENV1993-1.2 have been
evaluated and where appropriate implemented. A similar number of editorial comments have been
dealt with.
The conversion process has lead to various modifications with respect to the technical contents of
ENV1993-1.2. Many of these modifications were possible because, since the publication of ENV1993-1.2
in 1995, results of new, mainly European research, came available. Also a better harmonisation between
the various fire parts of Eurocode (concrete, steel, timber etc.) has been achieved.
The work on the conversion is nearly completed: a draft version of prEN1993-1.2 is available and will
shortly be discussed within Sub Committee 3. The formal voting is expected in the course of 2002.

ACKNOWLEDGEMENT

The author wishes to acknowledge the significant contributions of his fellow members in Project Team 2,
responsible for preparing the conversion of ENV1993-1.2 into an EN. Without these contributions the
conversion process would not have reached its present stage.

REFERENCES

[1]

[2]

[3]
[4]
[5]

ENV1993-1.2: Eurocode 3 Design of steel structures Part 1.2: General rules Structural fire
design.
CEN, September 1995.
prENV1993-1.2: Eurocode 3 Design of steel structures Part 1.2: General rules Structural
fire design, draft for voting.
CEN, September 2001.
Andersen, N.: Background for the deformation criteria and kx,.
SC3/PT2 doc. N024 (rev. 1), January 200.
Twilt, L.: Deformation criteria for fire exposed steelwork.
SC3/PT2 doc. N063(NL), November 2000.
prENV13381-4: Test method for determining the contribution to the fire resistance of structural
members: by applied protection to structural steel elements.

[6]

[7]

[8]

[9]

[10]

[11]

[12]
[13]
[14]
[15]

[16]
[17]

CEN TC127, september 2001.

Zhao, B.: Development of the use of stainless steel in construction; Work Package 5.1: Material
behaviour at elevated temperatures.
ECSC contract 7210-SA/842 etc., March 2000.
Latham, D.J. and Kirby, B.R.: Elevated Temperature Behaviour of Welded Joints in Structural
Steelwork.
ECSC contract 7210.SA/824 (F6.3/90).
Talamona, D., Franssen, J.M., Schleich, J.B. and Kruppa, J.: Stability of Steel Columns in Case
of Fire: Numerical Modelling.
Journal of Structural Engineering, June 1997.
Franssen, J.M., Talamona, D., Kruppa, J. and Cajot, L.G: Stability of Steel Columns in Case of
Fire: Experimental Evaluation.
Journal of Structural Engineering, February 1998.
Wainman, D.E. and Kirby, B.R.: Compendium of UK Standard Fire Test Data: Unprotected
Structural Steel 1 & 2.
British Steel Corporation, Swinden Laboratories, 1988.
Twilt, L. and Both, C.: Technical Notes on the Realistic Behaviour and Design of Fire Exposed
Steel and Composite Steel-Concrete Structures.
ECSC project SA112, 1991.
Kruppa, J., Twilt, L., Wesche, J. and Cooke, G.: Fire Protection of Structural Steelwork.
ECSC project SA 314, 1995.
Twilt, L.: Value for 1 to be taken into account in prEN1993-1.2.
SC3/PT2 doc. 085, April 2001.
Talamona, D.: Flambement de poteaux mttalique sous charges excentres haute temprature.
PhD thesis, Univ. Blaise Pascal, Clermont Ferrand, France, 1995.
ENV1993-1.1: Eurocode 3 Design of steel structures Part 1.1: General rules General rules
and rules for buildings.
CEN, 1995.
Leur, P.H.E. and Twilt, L.: Thermal Exposure in Fire Resistance Furnaces.
Proceedings of 6th IAFSS conference held in Poitier, June 5-9, 1999.
Twilt, L., Leur, P.H.E. v.d., and Both, C.: Characteristics of the heat transfer for calculating the
temperature development in structural steelwork exposed to standard fire conditions under plate
thermocouple control.
Proceedings of the first international workshop Structures in Fire, hold in Copenhagen June 19
and 20, 2001, September 2000.